摘要
为对开关柜绝缘状态进行评估,文中介绍了一种新型的关联规则数据挖掘(ARM)算法,用于识别开关柜局部放电的严重程度。该算法采用模糊C均值聚集(FCM)的方法划分局部放电特征量区间,基于改进型Apriori寻找满足最小支持度和最小可信度的候选集,对候选集进行递推和多次检索以产生用于分类的关联规则库。基于关联规则库对采集的10 kV开关柜中的多组针尖电晕局部放电信号进行模糊推理,结果表明采用关联规则的局部放电分类识别率高,分类结论准确,为开关柜绝缘状态评估提供了一定的理论依据和实际应用价值。
An improved association rules mining （ARM） algorithm is proposed for classification and recognition of partial discharge level to estimate the insulation status in switchgear cabinet. The method of fuzzy C-means clustering（FCM） is used to divide the range of partial Apriori algorithm is improved to find the candidate item sets which discharge characteristic amounts. The meet the minimum support and the minimum confidence. Recursion and retrieval are executed repeatedly to establish an association rule base for classification. A partial discharge experiment platform for 10 kV switchgear cabinet is set up, and several groups of corona partial discharge signals at different severity levels are detected for fuzzy reasoning. The results show that the signals can be classified into three severity levels accurately. The proposed classification method may benefit insulation status estimation and partial discharge detection.
出处
《高压电器》
CAS
CSCD
北大核心
2014年第2期23-28,共6页
High Voltage Apparatus
基金
国家自然科学基金资助项目(51307106)~~
